The Paleocene-Eocene Thermal Maximum or PETM

By Andy May

The PETM or Paleocene-Eocene Thermal Maximum was a warm period that began between 56.3 and 55.9 Ma (million years ago). The IPCC AR6 report (actually a draft, not a final edited report), released to the public on August 9, 2021, suggests that this warm period is similar to what is happening today and they expect to happen in the future (IPCC, 2021, pp. 2-82 & 5-14). During the PETM, it was very warm and average global surface temperatures probably peaked between 25.5°C and 26°C briefly, compared to a global surface temperature average of about 14.5°C today, as shown in Figure 1.

Figure 1. The estimated global average surface temperature for the past 150 million years. Modified from: Christopher Scotese, the Paleomap Project, link.

Today we have tens of thousands of daily temperature measurements around the world and can calculate a fairly accurate global average surface temperature. To construct a global average for the PETM we must rely on proxy temperatures, such as oxygen isotope ratios, Calcium/Magnesium ratios in fossil shells, and fossil membrane lipids that are sensitive to temperature like Tex86. Proxy temperature values are sparsely located and have a temporal resolution, 56 Ma, of thousands to hundreds of thousands of years. Thus, in terms of rate of temperature change, they are not comparable to today’s monthly global averages.

Before diving into the PETM, we will provide some geological perspective. According to Christopher Scotese, the highest global average temperature in the Phanerozoic (the age of complex shelled organisms, or the past 550 million years) was the Triassic hothouse event, following the end of the Karoo Ice Age, around 250-300 Ma. Global average surface temperatures peaked then at about 27.9°C.

The Late Cretaceous was the next warm period and global temperatures reached 24°C by 80 Ma. Around 65 Ma, the famous Chicxulub bolide impact occurred near the Yucatan Peninsula creating a 100 million megaton blast that ended the dinosaurs, except for birds. Global temperatures decreased rapidly to a minimum of around 12°C. This is a little colder than the last glacial maximum, so the animals that survived the blast got hit with the cold. This cold period was brief, perhaps only ten years, but very cold.

After recovering from the blast, the world returned to a global surface temperature of 22°C, just above its Phanerozoic average surface temperature of 20°C. Then, about nine million years later, the PETM warming occurred, it is labeled in Figure 1. In addition to the PETM, the EECO (Early-Eocene Climatic Optimum, mismarked in Scotese’s figure as EEOC), the MECO (Middle-Eocene Climatic Optimum), the EOT (Eocene-Oligocene Transition) Cooling, MMCO (Middle-Miocene Climatic Optimum), the LGM (last glacial maximum), and the preindustrial (Little Ice Age) temperature of 13.8° are identified in Figure 1. The current global average surface temperature of 14.5°C, and the possible “PAW” or projected anthropogenic global warming temperature of 19.8°C are also identified. Even if the most extreme projected anthropogenic warming occurs, it only takes Earth to its Phanerozoic average temperature of 20°C. This gives us an idea of just how historically, or geologically, cold it really is today.

Mixing proxy temperature resolutions of hundreds of thousands of years and modern instrumental temperature averages must be done with care. Scotese’s proxy temperatures are based on geological glacial signatures and estimated equator-to-pole temperature gradients. They are reasonable, but they have a temporal resolution of tens-of-millions of years. Given this resolution, we cannot compare past rates of warming, or cooling, to the past 170 years, the period with daily global thermometer coverage. His points do not have the temporal resolution or accuracy to allow for that.


Near the beginning of the EECO, there was an abrupt, but geologically short (< ~200,000 years) period of extreme warming usually called the “PETM” (Scotese, 2015). The PETM is the warmest of several short, very warm periods or “hyperthermals” that occurred in the late Paleocene and early Eocene between 56 and 53 Ma.

The PETM hyperthermal may have caused Earth’s surface to warm to nearly 26°C (79°F) at its peak, nearly 12°C warmer than today. At this time, SSTs (sea surface temperatures) reached 33°C in the North Atlantic, near Denmark (Stokke, Jones, Tierney, Svensen, & Whiteside, 2020). The PETM latitude of Denmark was about 45°N, south of its present latitude of 55.6°N, see Figure 3 for its location at the time. The warming is contemporaneous with a small-scale marine extinction event. The high temperatures also occur at about the same time as a carbon isotopic excursion (CIE) event. This means the carbon-13 and carbon-12 ratio changed abruptly for a short time, geologically. The CIE suggests a flow of carbon compounds, enriched in carbon-12, into the ocean and atmosphere. Hydrocarbons are enriched in carbon-12 because plants prefer it over carbon-13, so the source could be hydrocarbons, such as methane clathrates. Clathrates are compounds in which molecules of one component, in this case methane, are trapped in another, like water. Methane clathrates are sometimes called methane hydrates.

Figure 2. The section studied by Stokke, Fur Island, Denmark. Figure modified after (Stokke, Jones, Tierney, Svensen, & Whiteside, 2020).

Figure 2 shows the section studied by Stokke, et al. There is an unconformity below the section, but the PETM deposit is thought to be complete. Sedimentation was very rapid and the entire 24-meter PETM section was probably deposited in 100,000 years. NAIP (North Atlantic Igneous Province, see Figure 3) ash deposits within the PETM body are dated to 55.6 Ma. Contemporaneously ocean pH and deep-ocean oxygen levels lowered (anoxia) (IPCC, 2021, p 2-82).

It is only barely visible in Figure 2, but after the unconformity, and just prior to the PETM warming event, North Sea SSTs drop to 14.5°C, before rising rapidly to 33°C. Today’s global, area-weighted, average SST is about 18.3°C, using NOAA MIMOC and University of Hamburg multiyear ocean mixed layer temperature data. Today’s average North Sea SST is about 11°C according to, so the temperature of 14.5°C prior to the PETM is a little warmer than today. According to NOAA MIMOC data, the global average SST, in the mixed layer, at 45°N is about 10°C today.

NAIP volcanism was active during and before the PETM. Basalts, five kilometers thick, were emplaced on eastern Greenland and the Faroe Islands between 56 and 55.6 Ma. This pre-PETM volcanic episode would have generated sulfur compounds. The sulfur compounds (such as SO2 and SO3), when combined with water, generate sulfuric acid, which cooled the atmosphere and oceans, lowered ocean pH, and removed oxygen from both the atmosphere and oceans.

Francesca McInerney and Scott Wing have also written about the PETM (McInerney & Wing, 2011). They emphasize the global impact of the event. They believe that global surface temperature increased 5-8°C. While Stokke, et al. place the beginning of the PETM at 55.9 Ma, McInerney and Wing place it at 56.0 to 56.3 Ma. The time difference is not significant, given the dating uncertainty.

The data suggests that the PETM extinctions were quite limited and coincident with a large increase in mammalian species. Approximately 30-50% of benthic (bottom dwelling) foraminifera and dinoflagellates (microscopic marine animals) went extinct. The benthic foraminifera that went extinct were mostly from the middle to deeper depths in the oceans. Some speculate that the extinctions were due to greater corrosivity of deeper waters, lower oxygen levels, and/or higher temperatures (McInerney & Wing, 2011). Other benthic animals, such as ostracodes, living in the same environment as the foraminifera did not show the same loss in numbers. McInerney and Wing speculate that the extinctions were mostly due to higher temperatures, but this is unclear. Temperatures were higher at the surface and on land, yet life thrived in these environments at the time.

Oddly, while benthic forams did not do well during the PETM, McInerney and Wing report their planktonic (floating) cousins did very well, increasing in both size and diversity. Some planktonic species greatly increased their range and numbers during the PETM.

On land, new mammals, especially primates, evolved at this time and spread widely. The PETM and EECO saw a “burst of mammalian first appearances.” The period is sometimes called the “mammalian dispersal event” (McInerney & Wing, 2011). This burst of new mammalian first appearances is seen in both North America and Europe, and coincides with the onset of the PETM CIE.

Besides mammals, the PETM and EECO saw the evolution and dispersal of numerous new and existing species of turtles and lizards. Suggestions that the PETM was so warm it was deadly in the tropics are unlikely to be true “because terrestrial floras from the tropics become more rather than less diverse during the PETM” (McInerney & Wing, 2011).

During the PETM, temperatures increased 6 to 12°C and there was no polar ice. Palm trees grew in the Arctic, and Antarctica was covered in forests. Biological diversity greatly increased during the PETM, especially among terrestrial plants (McInerney & Wing, 2011). Some plant species appeared to disappear during the PETM but reappeared later, suggesting they did not extirpate, but were simply not preserved for a short time. Many species flourished and expanded into new areas. The first occurrence of mangrove palms dates to the PETM. Overall, nature blossomed when global temperatures were likely 12°C warmer than today.

The cause of the PETM is unknown. As mentioned above, Ella Stokke’s work suggests that the warm period and the extinctions are closely related to, and possibly caused by, contemporaneous North Atlantic Igneous Province (NAIP) volcanism. The volcanism could have caused oxygen deprivation in the Atlantic Ocean, especially in the deep Atlantic, causing the benthic foraminifera extinctions. The NAIP region and the Fur Island, Denmark, area studied by Stokke and her colleagues are shown in Figure 3. The Fur Island outcrops provide a fairly complete geological record spanning the Paleocene-Eocene boundary. The sediments include volcanic ash beds from NAIP, fossil-rich claystones, and shales with clear records of the carbon isotope excursion and the PETM event. They used the TEX86 temperature proxy to estimate SST. The NAIP emplacement was most active between 56 and 54 Ma during the opening of the North Atlantic.

Figure 3. The North Atlantic Igneous Province and Stokke’s study area. Modified from: E.A.M. van de Lisdonk, Utrecht University.

Many theories for the warming have been proposed, including a sudden release of CO2 and methane, but the evidence is ambiguous. Initially it was speculated that volcanism caused the release of a large amount of methane clathrates, which then caused temperatures to rise.

A lot of carbon dioxide and methane certainly entered the atmosphere during this time. The main problem with the methane clathrate theory is that there were not enough of them to supply the necessary carbon (McInerney & Wing, 2011). Other possible sources of the excess carbon in the atmosphere and oceans at the time, enriched in carbon-12, are volcanism and contact metamorphism in the NAIP (Stokke, Jones, Tierney, Svensen, & Whiteside, 2020). McInerney and Wing prefer the theory that Antarctic peat and permafrost melted and were the source of the excess carbon-12.

While total atmospheric carbon did increase in the PETM, climate models have not been able to reproduce the large temperature increase with reasonable parameters. The CO2 estimates (see Figure 4) that we have from the period are far too low. Even if the atmosphere is assumed to contain 16 times the preindustrial concentration of CO2 (4,800 ppm) many times the level suggested by the fossil evidence shown in Figure 4, and the climate sensitivity is assumed to be 3°C/2xCO2 (McInerney & Wing, 2011), the CO2 forcing is still not enough to cause the warming observed in the sedimentary record (IPCC, 2021, 5-14). The abbreviation “°C/2xCO2” means the rise in temperature due to doubling the CO2 atmospheric concentration.

NASA claims their simulations can model the PETM temperature rise if they incorporate very high CO2 sensitivity. Jiang Zhu and colleagues successfully simulated the PETM with existing data, but their model suggests a climate sensitivity of 6.6°C/2xCO2, which is not reasonable (Zhu, Poulsen, & Tierney, 2019). The IPCC AR6 very likely range of sensitivity is 2°C to 5°C/2xCO2 (IPCC, 2021, pp. TS-58). In AR5, the IPCC is more explicit: “ECS is very unlikely greater than 6°C is an expert judgment informed by several lines of evidence.” (IPCC, 2013, p. 1111). AR5 follows with a list of the evidence why ECS is not greater than 6.

Figure 4. Stokke’s Danish Sea surface temperatures (SSTs) are shown in blue and Beerling’s CO2 estimates are shown with orange boxes.

Between 55 and 56 Ma, there are 16 proxy estimates of CO2 levels (Beerling & Royer, 2011, S.I. Table 1). The 16 proxy samples represent seven unique sample times, which are plotted in Figure 4. According to a summary paper by David Beerling and Dana Royer in Nature Geoscience, the CO2 concentration was about 487 ppm (328-667 ppm) during the PETM event. The dating error for the samples is estimated to be ±500,000 years and the error in the CO2 estimates is ±235 ppm (Beerling & Royer, 2011). There are four methods of determining ancient CO2 atmospheric concentrations according to Beerling and Royer. The two terrestrial methods are the abundance of stomatal pores on fossil leaves and the carbon isotope composition of carbonates in fossil soils. Leaf stomata reduce in number when CO2 becomes abundant and gain in number when it is lower, as it is today.

The two marine methods are the carbon isotope composition of phytoplankton fossils and the boron isotope composition of fossil foraminifera. I refer you to Beerling and Royer’s paper and their references for the details of how these methods work. There is also a decent summary in AR6 (IPCC, 2021, 2-15). The general uncertainty in CO2 concentration estimates in the Eocene is nearly 100%, that is the actual value is between twice the most likely value and half of it.

Stokke’s Denmark SSTs are also plotted in Figure 4 for comparison. The CO2 estimates are somewhat lower than other PETM estimates and only slightly higher than today, but global average temperatures were 10-12°C higher.

Today we measure atmospheric CO2 continuously, many times a day, and each sample has a precise date and time. The data from 56 Ma is obviously not comparable to today. Beerling and Royer claim the highest CO2 level occurred 52 Ma, over four million years after the PETM, see Figure 5. They have two CO2 estimates from that time period: 1,868 ppm (1,092 – 3.501 ppm) and 659 ppm (439 – 878 ppm). Elevated CO2 levels existed from 54 to 32 Ma and average around 800 ppm. The lack of correlation between increasing CO2 and temperature from the period 52 to 57 Ma is easily seen in Figure 4.

Figure 5. Various estimates of CO2 concentration in the Cenozoic. After: (Beerling & Royer, 2011).

During the PETM, we can be confident that CO2 levels were only slightly higher than today, not high enough to be a significant factor in the warming event, as acknowledged in AR6 (IPCC, 2021, p 5-14). We certainly do not have any idea about the rate of PETM warming or the rate of CO2 increase, relative to today, since the estimates are thousands to hundreds of thousands of years apart.

The IPCC wants to use the PETM as an example of what can happen today, but they admit low to medium confidence in the amount of carbon released during the PETM and their inferred increase in CO2 can only account for half the estimated warming during that time (IPCC, 2021, p. 5-14). They are also trying, without success so far, to model global temperatures in the PETM. Their models of the CO2 impact on climate in the PETM and other selected periods do not predict the warming observed, so one could reasonably conclude that this means the models are not working. But, in AR6, they conclude that it means the feedbacks to surface temperature are changing with surface temperature. Thus the feedbacks have feedbacks (IPCC, 2021, p. 7-78). They do not believe the models can be incorrect, they conclude that we simply need to introduce another factor. This is Karl Popper’s definition of pseudoscience, a hypothesis that cannot be falsified. The PETM is an interesting time in geological history, but the causes of the warming, the lower ocean pH, the limited ocean extinctions, and increasing mammalian diversity are unclear. One thing for sure, the PETM is not an analogue for today.

The bibliography can be downloaded here.